LAUSR

The complexity and variability of mobile social networks make protocol evaluation hard. Thus synthetic mobility models that well reflect the properties of human movement in real MSNs must be used in simulations. The overall objective of this paper is to design a pragmatic mobility model that comprehensively involves multiple factors that affect the choice of the next destination. The concept of Community Attraction is proposed as the selection criteria. It is related to three factors, that is, the distance of moving, the human relationships and the location restriction. Thus, our new mobility model is called Distance, Relationship, Location (DRL). Specifically, the former two factors are indicated through interaction matrices, which take the Social Relationship Attributes and the information of location as input. And we propose Location Attraction for the first time to denote the location restriction of a place. By the way, the value of Location Attraction is time varying. Moreover, the parameters that decide the weights of the factors in the formula of Community Attraction are derived by machine learning. And the learning method is called Bayesian Personalized Ranking algorithm. We load several protocols on DRL and the result shows that DRL correctly assesses their performance. To verify the reasonability of our model, we compare the simulation results of DRL with real traces, and they fit well.